In attempting to bisect to identify a regression range, I found that this doesn't reproduce at all using our APKs from mozregression (i.e. APKs downloaded from archive.mozilla.org). Even if I use these APKs for 143.0.3, I don't hit the bug: https://archive.mozilla.org/pub/fenix/releases/143.0.3/android/fenix-143.0.3-android/fenix-143.0.3.multi.android-universal.apk https://archive.mozilla.org/pub/fenix/releases/143.0.3/android/fenix-143.0.3-android-armeabi-v7a/fenix-143.0.3.multi.android-armeabi-v7a.apk ...and yet 143.0.3 installed from the Play store does hit the bug. However: tcampbell shared instructions for how to generate and install the same sort of collection-of-installer-packages that we use on the Play Store from archive.mozilla.org's "aab" files, and with that, I *am* able to reproduce the bug (using APKs generated locally from aab files downloaded from archive.mozilla.org). That let me bisect to this regression range: https://hg-edge.mozilla.org/mozilla-central/pushloghtml?fromchange=ebf908f86af1&tochange=2c28b9cc7958 * Last known-good build (plays my test YouTube video just fine): https://archive.mozilla.org/pub/fenix/nightly/2025/07/2025-07-23-21-57-55-fenix-143.0a1-android/fenix-143.0a1.multi.android.aab Built from hg revision ebf908f86af1 * First known-bad build (hits the YouTube error described in this bug): https://archive.mozilla.org/pub/fenix/nightly/2025/07/2025-07-24-21-29-06-fenix-143.0a1-android/fenix-143.0a1.multi.android.aab Built from hg revision 2c28b9cc7958 Pushlog: https://hg-edge.mozilla.org/mozilla-central/pushloghtml?fromchange=ebf908f86af1&tochange=2c28b9cc7958 In that range, we have this commit which I'm tending to suspect as the regressor: https://hg-edge.mozilla.org/mozilla-central/rev/e8aefd74680cef2bc0ddbd3967102308f06ce214 > Bug 1867280 - Stop using legacy packaging for native libraries. It's suspicious because: (a) it's about packaging and has several comments mentioning aab files, (b) it's about "legacy" Androids stuff, which seems relevant given that this bug is about Android 6, and (c) it was specifically removing a legacy thing, and it seems conceivable that maybe older Android versions might depend on legacy stuff. RyanVM, could we try reverting that patch? If we can somehow generate .aab builds for the 143 release repo before/after we hypothetically revert that patch, I'm happy to test them to confirm that it fixes the issue.
Bug 1992172 Comment 11 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
In attempting to bisect to identify a regression range, I found that this doesn't reproduce at all using our APKs from mozregression (i.e. APKs downloaded from archive.mozilla.org). Even if I use these APKs for 143.0.3, I don't hit the bug: https://archive.mozilla.org/pub/fenix/releases/143.0.3/android/fenix-143.0.3-android/fenix-143.0.3.multi.android-universal.apk https://archive.mozilla.org/pub/fenix/releases/143.0.3/android/fenix-143.0.3-android-armeabi-v7a/fenix-143.0.3.multi.android-armeabi-v7a.apk ...and yet 143.0.3 installed from the Play store does hit the bug. However: tcampbell shared instructions for how to generate and install the same sort of collection-of-installer-packages that we use on the Play Store from archive.mozilla.org's "aab" files, and with that, I *am* able to reproduce the bug (using APKs generated locally from aab files downloaded from archive.mozilla.org). That let me bisect to this regression range: https://hg-edge.mozilla.org/mozilla-central/pushloghtml?fromchange=ebf908f86af1&tochange=2c28b9cc7958 * Last known-good build (plays my test YouTube video just fine): https://archive.mozilla.org/pub/fenix/nightly/2025/07/2025-07-23-21-57-55-fenix-143.0a1-android/fenix-143.0a1.multi.android.aab Built from hg revision ebf908f86af1 * First known-bad build (hits the YouTube error described in this bug): https://archive.mozilla.org/pub/fenix/nightly/2025/07/2025-07-24-21-29-06-fenix-143.0a1-android/fenix-143.0a1.multi.android.aab Built from hg revision 2c28b9cc7958 Pushlog: https://hg-edge.mozilla.org/mozilla-central/pushloghtml?fromchange=ebf908f86af1&tochange=2c28b9cc7958 In that range, we have this commit which I'm tending to suspect as the regressor: https://hg-edge.mozilla.org/mozilla-central/rev/e8aefd74680cef2bc0ddbd3967102308f06ce214 > Bug 1867280 - Stop using legacy packaging for native libraries. It's suspicious because: (a) it's about packaging and has several comments mentioning aab files, (b) it's about "legacy" Androids stuff, which seems relevant given that this bug is about Android 6, and (c) it was specifically removing a legacy thing, and it seems conceivable that maybe older Android versions might depend on legacy stuff. (And I wouldn't be surprised if we're using a native library for our audio decoding.) RyanVM, could we try reverting that patch? If we can somehow generate .aab builds for the 143 release repo before/after we hypothetically revert that patch, I'm happy to test them to confirm that it fixes the issue.
In attempting to bisect to identify a regression range, I found that this doesn't reproduce at all using our APKs from mozregression (i.e. APKs downloaded from archive.mozilla.org). Even if I use these APKs for 143.0.3, I don't hit the bug: https://archive.mozilla.org/pub/fenix/releases/143.0.3/android/fenix-143.0.3-android/fenix-143.0.3.multi.android-universal.apk https://archive.mozilla.org/pub/fenix/releases/143.0.3/android/fenix-143.0.3-android-armeabi-v7a/fenix-143.0.3.multi.android-armeabi-v7a.apk ...and yet 143.0.3 installed from the Play store does hit the bug. However: tcampbell shared instructions for how to generate and install the same sort of collection-of-installer-packages that we use on the Play Store from archive.mozilla.org's "aab" files, and with that, I *am* able to reproduce the bug (using APKs generated locally from aab files downloaded from archive.mozilla.org). That let me bisect to this regression range: https://hg-edge.mozilla.org/mozilla-central/pushloghtml?fromchange=ebf908f86af1&tochange=2c28b9cc7958 * Last known-good build (plays my test YouTube video just fine): https://archive.mozilla.org/pub/fenix/nightly/2025/07/2025-07-23-21-57-55-fenix-143.0a1-android/fenix-143.0a1.multi.android.aab Built from hg revision ebf908f86af1 * First known-bad build (hits the YouTube error described in this bug): https://archive.mozilla.org/pub/fenix/nightly/2025/07/2025-07-24-21-29-06-fenix-143.0a1-android/fenix-143.0a1.multi.android.aab Built from hg revision 2c28b9cc7958 Pushlog: https://hg-edge.mozilla.org/mozilla-central/pushloghtml?fromchange=ebf908f86af1&tochange=2c28b9cc7958 In that range, we have this commit which I'm tending to suspect as the regressor: https://hg-edge.mozilla.org/mozilla-central/rev/e8aefd74680cef2bc0ddbd3967102308f06ce214 > Bug 1867280 - Stop using legacy packaging for native libraries. It's suspicious because: (a) it's about packaging and has several comments mentioning aab files, (b) it's about "legacy" Androids stuff, which seems relevant given that this bug is about Android 6, and (c) it was specifically removing a legacy thing, and it seems conceivable that maybe older Android versions might depend on legacy stuff. (And I wouldn't be surprised if we're using a native library for our audio decoding.) RyanVM, could we try reverting that patch? If we can somehow generate .aab fenix builds for the 143 release repo before/after we hypothetically revert that patch, I'm happy to test them to confirm that it fixes the issue.